Factored Agents: Decoupling In-Context Learning and Memorization for Robust Tool Use
This addresses robustness issues in agentic AI systems, such as malformed API fields and suboptimal planning, though it appears incremental as it builds on existing LLM-based designs.
The paper tackles the limitations of single-agent systems in agentic AI by proposing a factored agent architecture that decouples in-context learning and memorization, resulting in significant improvements in planning accuracy and error resilience.
In this paper, we propose a novel factored agent architecture designed to overcome the limitations of traditional single-agent systems in agentic AI. Our approach decomposes the agent into two specialized components: (1) a large language model (LLM) that serves as a high level planner and in-context learner, which may use dynamically available information in user prompts, (2) a smaller language model which acts as a memorizer of tool format and output. This decoupling addresses prevalent issues in monolithic designs, including malformed, missing, and hallucinated API fields, as well as suboptimal planning in dynamic environments. Empirical evaluations demonstrate that our factored architecture significantly improves planning accuracy and error resilience, while elucidating the inherent trade-off between in-context learning and static memorization. These findings suggest that a factored approach is a promising pathway for developing more robust and adaptable agentic AI systems.